CasualtyBase
Description
Historical Database / Web App
October 2026
CasualtyBase makes 6.5 million U.S. Army hospital admission cards from 1942–1945 searchable by service number. Each card was punched in code for the Army’s Office of the Surgeon General, later copied to magnetic tape, and is now released by the National Archives (Record Group 112) as raw lines of digits that are unreadable without the original code books.
An ETL pipeline cuts every line into fields using the National Archives’ layout charts, since the Army used 19 different card layouts over the four years. It translates each code with the Army’s own code lists and the decoding programs written by the National Personnel Records Center in 1988, then maps every year onto one set of plain-English headings: when, where, diagnosis, cause and outcome.
All cards with the same service number appear on one timeline, with every coded field shown next to its decoded meaning. Automated tests compare the output with real cards and check the results against common sense, and documented rules cover the judgement calls, such as estimated discharge dates, deaths on early cards and near-miss service numbers.
A statistics page lets visitors filter all 6.5 million cards by year, type of case, outcome and place, with interactive charts.
Technology
Python
Flask- ETL pipeline
- Automated testing
- Interactive charts